Objective We aimed to compare a combined oral contraceptive (COC) plus the antiandrogen spironolactone with the insulin sensitizer metformin in women with polycystic ovary syndrome (PCOS). Design We conducted a randomized, parallel, open-label, clinical trial comparing COC (30 μg of ethinylestradiol and 150 μg of desogestrel) plus spironolactone (100 mg/day) with metformin (850 mg b.i.d.) for one year in women with PCOS (EudraCT2008–004531–38). Methods The composite primary outcome included efficacy (amelioration of hirsutism, androgen excess and menstrual dysfunction) and cardiometabolic safety (changes in the frequencies of disorders of glucose tolerance, dyslipidemia and hypertension). A complete anthropometric, biochemical, hormonal and metabolic evaluation was conducted every three months and data were submitted to intention-to-treat analyses. Results Twenty-four patients were assigned to COC plus spironolactone and 22 patients to metformin. Compared with metformin, COC plus spironolactone caused larger decreases in hirsutism score (mean difference 4.6 points, 95% CI: 2.6–6.7), total testosterone (1.1 nmol/L, 0.4–1.7), free testosterone (25 pmol/L, 12–39), androstenedione (5.5 nmol/L, 1.8–9.2) and dehydroepiandrosterone sulfate (2.7 μmol/L, 1.4–4.0). Menstrual dysfunction was less frequent with COC plus spironolactone (OR: 0.06, 95% CI: 0.02–0.23). No differences were found in frequencies of abnormal glucose tolerance (OR: 1.7, 95% CI: 0.7–4.4), dyslipidemia (OR: 0.6, 95% CI: 0.2–1.8) or hypertension (OR: 0.3, 95% CI: 0.5–2.0). No major adverse events occurred and biochemical markers were similarly safe with both treatments. Conclusions COC plus spironolactone was more effective than metformin for symptoms of PCOS showing similar safety and overall neutral effects on cardiometabolic risk factors.